| Summary: | 碩士 === 國立臺北科技大學 === 電機工程研究所 === 104 === Aerobics is one of the best exercises for a full-body or functional workout. However, flexibility of aerobics also makes it difficult to estimate calories consumption. The single accelerometer method failed to provide an accurate model due to complicated movements of aerobics. This thesis proposed a method to accurately estimate the calories consumption.
The proposed method took advantage of Microsoft Kinect to track 10 body joints of exercisers. Each node could be considered as a 3-axis accelerometer mounting the corresponding joint. Three regression models were exploited to form estimation models, which included the linear, multiple regression model, and support vector regression. The accuracy and robustness of the proposed estimation models were evaluated in the experimental studies. The experiments analyzed three models by 30, 20, and 10 minutes of exercises. The average error rate of estimating calories consumption is 3.48%, 3.66% and 2.72% for each corresponding period. This thesis concluded that the best estimation model came from support vector regression. The contributions could help to build a support vector regression estimation system for accurate and robust calories consumption of aerobics.
|
|---|
